Skip to main content

Full text of "Novel Transport Mechanism for Interacting Electrons in Disordered Systems: Variable-Range Resonant Tunneling"

See other formats


Novel Transport Mechanism for Interacting Electrons 
in Disordered Systems: 
Variable- Range Resonant Tunneling 



S.D. Baranovskii 1 and I.S. Shlimak 2 



1 Institute for Physical Chemistry and Material Sciences Center, 
Philipps-University Marburg, D-35032 Marburg, Germany 
2 Jack and Pearl Resnick Institute of Advanced Technology, 
Department of Physics, Bar-Ilan University, Ramat-Gan 52900, Israel 

To interpret recent experimental observations of the phononless hopping con- 
duction, we suggest a novel transport mechanism according to which the current- 
carrying single electrons move via quantum resonant tunneling between localized 
states brought into resonance by fast electron hops in their environment. 



Recent experiments on the quantum Hall effect [1,2] and on the hopping 
transport in 2D Si MOSFETs [3] and delta-doped GaAs/AlGaAs heterostruc- 
tures [4] show that prefactor so in the exponential temperature dependence 
of the hopping conductivity has, under certain conditions, a universal form 
(e 2 /h)f(T/T Q ) with / «1 at T w T . Aleiner et al. [5] have remarked 
that such an effect can be due to hopping via electron-electron scattering, 
as considered by Fleishman et al. [6], rather than via conventional electron- 
phonon scattering. We suggest below an alternative to this mechanism based 
on non-Hubbard Coulomb correlations which can account for the observed 
effects. 

To interpret phononless hopping evidenced by experimental observations, 
it is necessary to take into account electron-electron interactions, though such 
interactions can also play an important role for phonon-assisted hopping. The 
role of non-Hubbard Coulomb correlations for the low-temperature hopping 
conduction was first pointed out by Knotek and Pollak [7]. They considered 



1 



two types of such correlations which are caused by interaction between elec- 
trons on different sites: sequential correlations (SC) and electron polarons 
(EP). In the case of SC, the single-electron current-determining hops can 
be assisted by preliminary hops between surrounding states. These prelimi- 
nary hops prepare the favorable situation for the current-carrying transitions 
of single electrons. In the case of EP, several electrons hop simultaneously. 
Such simultaneous multielectron transitions can have lower activation en- 
ergies than single-electron hops and therefore they can be efficient [7]. In 
contrast to the case of the sequential transitions, the probability for a si- 
multaneous transition of several electrons involves the product of the fac- 
tors exp{— 2r/a} for each participating single-electron hop (where r is the 
hop distance and a is the localization length). Therefore, such transitions 
can be neglected when the concentration of localized states is sufficiently 
low, although for intermediate doping levels such transitions can be efficient 
[8]. Pollak and his coworkers have also emphasized the importance of quan- 
tum effects in disordered systems of localized electrons interacting via their 
Coulomb potentials [9,10]. By computer simulation they studied the level 
spacing spectra as a measure of the degree of localization, though not di- 
rectly addressing the conduction phenomena. 

The question most widely discussed related the electron-electron correla- 
tions is whether or not the temperature dependences of the hopping conduc- 
tivity obtained for single-electron hopping are altered by taking into account 
the many-particle effects. Perez-Garrido et al. [11] studied this problem by 
computer simulation and concluded that while both correlation effects have a 
significant influence on the hopping conductivity, its temperature dependence 
at low T is described quite well by the dependence predicted by theories of 
single-electron hopping, although with changed numerical parameters. In the 
Coulomb gap regime, the Efros-Shklovskii formula should be valid [8] 

aaexpl-^/T) 1 / 2 }, (1) 
whereas at higher temperatures, the Mott formula applies: 

acxexpj-^/T) 1 /^ 1 )}, (2) 

where d is the spatial dimensionality and T 1; T 2 are characteristic tempera- 
tures. 



2 



So far, most research has been concentrated on the phonon-assisted hop- 
ping processes. Only one paper [6] has been devoted to the temperature 
dependence of the hopping conductivity in the case of phononless hopping. 
Fleishman et al. [6] have studied EP correlations, i.e., simultaneous transi- 
tions of several electrons in the phononless mode. They succeeded in showing 
that for the phononless transport of multielectron excitations, the temper- 
ature dependence of the conductivity is described by Eq.(2) in the Mott 
regime. It may well be the proposed by Aleiner et al. [5], that it is 

this transport mechanism, which is responsible for the recent observations of 
the phononless hopping in 2D systems. Not arguing in favour or against this 
hypothesis, we suggest below an alternative mechanism of the phononless 
hopping processes which also provides the observed temperature dependen- 
cies of the hopping conductivity. We do not aim at calculation of prefactors 
and concentrate our discussion on exponential terms. 

We assume that the current-carrying hop between the sites belonging to 
the percolation cluster is represented by a single-electron resonant transition, 
the resonance being prepared by fast assisting hops in the environment of 
these sites. It is a kind of SC with the current- carrying hop provided by 
resonant tunneling. At first glance, this looks similar to the picture recently 
discussed by Agrinskaya and Kozub [12]. They also considered SC and they 
also assumed assisting hops to be fast compared to the current-carrying hop. 
However, Agrinskaya and Kozub suggested for the rate W of the current- 
carrying hop expression 

W oc exp{— 2r/a — e/kT}, (3) 

where the activation term describes the probability of an assisting hop, and r 
is the distance between the corresponding pair of the current-carrying sites. 
The activation energy of the current-carrying hop was assumed to vanish due 
to averaging over fast assisting hops [12]. As long as the assisting hops are 
due to electron-phonon scattering, the phonon frequency is inherently present 
in the rate of current-carrying hops of Agrinskaya and Kozub [12]. We argue 
below that in our mechanism of resonant transitions, the rate W of a current- 
carrying hop does not involve the phonon frequency even if the assisting hops 
are due to the interaction with phonons. 

Let us consider two neighboring sites i and j belonging to the percola- 
tion cluster, so that an electron transition between them is necessary for 
dc transport. We assume that because of fast electron transitions in the 



3 



sites surrounding i and j, the electron energies £j and £j on the chosen sites 
fluctuate with some typical frequency u due to the Coulomb potentials of 
surrounding electrons. In fact, it is sufficient to consider the fluctuations of 
just one of the two energies e« or Sj. Due to these electron correlations, a 
resonant situation 

\e i -e J \<I ij (4) 

can be established for some short time interval ti, where Iij oc exp(— r^/a) 
is the overlap integral between sites % and j. If the amplitude of the energy 
fluctuations is E » 1^, this time interval can be approximated by 

tj » Iij/iEu) (5) 

The spreading out of the electron wave function between the resonant sites 
obeying condition (4) takes a time r of order of h/Iy. If the energy fluc- 
tuations are fast compared to r, i.e., ujt » 1, then the probability of the 
transition occuring during one fluctuation period is tj/t and the transition 
rate is 

W « wij/r (6) 

Using Eq.(5), one obtains 

W » 11/ {Eh) (7) 

This formula shows that the frequency uj of the assisting hops is not 
involved in the hopping rate for the current- carrying hops if the condition 
ujt » 1 is fulfilled, i.e., if the assisting hops are much faster than the current- 
carrying hop. This condition is the main requirement of all SC models. 

Above we considered the situation in which the current-carrying pair of 
sites i and j already has one electron. In order to satisfy this condition, one 
should take into account the energy distribution function of the electrons 
f(e). Therefore, the phononless electron transition rate i/^ from the site % to 
the site j has the form 

u ij = Wf(e i )[l-f(e j )] (8) 



4 



where in the Ohmic regime, f(e) should be replaced by the Fermi equilibrium 
distribution [8]. This leads to the expression 

Uij = u exp{-2rij/a}exp{-eij/kT} (9) 
for arbitrary positions of £j and Ej with respect to the Fermi level \x. Here 
Eij = (l/2){£j - €j+ | Si - 11 | + | £j - n |} (10) 

It is worth noting that our expression for differs slightly from the 
conventional one considered in hopping conduction, not being determined by 
the phonon characteristics. 

In order to calculate the exponential terms in the variable-range hop- 
ping (VRH) conductivity using the phononless transition rates determined 
by Eqs.(9) and (10), one can now repeat the routine derivation described in 
detail in handbooks, e.g. Ref. [8]. This derivation automatically leads to the 
dependences (1) and (2) [8]. All phonon-dependent factors in our approach 
can appear only in terms of the quantity u, which is not present in the fi- 
nal result for the hopping rate (Eq.(9)). The suggested model of electron 
transport joins together a resonant tunneling mechanism of band conductiv- 
ity with the necessity of finding an electron in the initial state and an empty 
final state, which is peculiar for the variable-range hopping conductivity and 
results in the temperature dependence of this band-like electron motion. It is 
therefore not surprising that experiments [3,4] gives evidence for the depen- 
dences (1) and (2) in the phononless VRH regime. We call our mechanism 
Variable-Range Resonant Tunneling. 

At the present stage, it is not possible to determine by comparison with 
experiment which particular phononless transport mechanism, - that of Ref. 6 
or the one suggested here - is responsible for the observed effects. Both mech- 
anisms lead to the same temperature dependences of the conductivity. An- 
other problem is to find the conditions for which phononless hopping becomes 
more efficient than phonon-assisted hopping. The rarity of the observations 
of the phononless VRH justifies, in most cases, the suggestion that electron- 
phonon scattering is responsible for VRH transport. However, in the vicinity 
of the metal-insulator transition, quantum effects should become pronounced 
[9], and under such conditions, the phononless mechanism can become domi- 
nant. Indeed, recent observations of phononless VRH [4] have been made on 
the system close to the metal-insulator transition. 



5 



In conclusions, one can claim the following. A novel transport mechanism 
is suggested for conduction in disordered systems with localized electrons 
interacting via their Coulomb potentials. It is based on the resonant single- 
electron tunneling between current-carrying sites brought into resonance by 
fast electron hops in their surrounding. These assisting hops can be due to 
electron-phonon scattering. If the typical rate of these transitions is much 
higher than that of the current-carrying hops, the former does not influence 
the result for the temperature dependence of VRH conductivity. 

We thank M. Pollak and A. Mobius for valuable discussions, in clarifying 
the results of Ref.6 and Ref.10. Financial support of the Deutscheforschungs- 
gemeinschaft via SFB 383 and of the Fonds der Chemischen Industrie is 
gratefully acknowledged. 



6 



References 

[I] G. Ebert, K. von Klitzing, C. Probst, E. Schubert, K. Ploog, and G. 
Weimann, Solid State Commun. 45, 625 (1983). 

[2] A. Briggs, Y. Guldner, J.P. Vieren, M. Voos, J.P Hirtz, and M. 
Razeghi, Phys. Rev. B 27, 6549 (1983). 

[3] W. Mason, S.V. Kravchenko, G.E. Bowker, and J.E. Furneaux, Phys. 
Rev. B 52, 7857 (1995). 

[4] S.I. Khondaker, IS. Shlimak, J.T. Nicholls, M. Pepper, and D.A. 
Ritchie, in Proceedings of the 24th International Conference on the Physics 
of Semiconductors, Jerusalem , 1998, edited by M. Heiblum and E. Cohen 
(World Scientific, Singapore, 1998), in press; Phys. Rev. B, in press. 

[5] I.L. Aleiner, D.G. Polyakov, and B.I. Shklovskii, in Proceedings of the 
22nd International Conference on the Physics of Semiconductors, Vancouver, 
1994, edited by D.S. Lockwood (World Scientific, Singapore, 1994), p. 787. 

[6] L. Fleishman, D.C. Licciardello, and PW. Anderson, Phys. Rev. Lett. 
40, 1340 (1978). 

[7] M.L. Knotek and M. Pollak, J. Non-Cryst. Solids 8-10, 505 (1972); 
Phys. Rev. B 9, 644 (1974). 

[8] B.I. Shklovskii and A.L. Efros, "Electronic Properties of Doped Semi- 
conductors" (Springer, Berlin, 1984) p. 251. 

[9] J. Talamantes, M. Pollak, and L. Elam, Europhys. Lett. 35, 511 
(1996). 

[10] J. Talamantes and A. Mobius, phys. stat. solidi (b) 205, 45 (1998). 

[II] A. Perez- Garrido, M. Ortuno, E. Cuevas, J. Ruiz, and M. Pollak, 
Phys. Rev. B 55, R8630 (1997). 

[12] N.V. Agrinskaya and V.I. Kozub, Semiconductors 32, 631 (1998). 



7